林业科学 ›› 2026, Vol. 62 ›› Issue (11): 187-196.doi: 10.11707/j.1001-7488.LYKX20260089
• 研究论文 • 上一篇
收稿日期:2026-02-09
修回日期:2026-09-21
出版日期:2026-11-10
发布日期:2026-11-20
通讯作者:
王忠伟
E-mail:wangpmp@163.com
基金资助:
Chen Long1,2,Zhongwei Wang2,*(
),Qingtian Wang3
Received:2026-02-09
Revised:2026-09-21
Online:2026-11-10
Published:2026-11-20
Contact:
Zhongwei Wang
E-mail:wangpmp@163.com
摘要:
目的: 针对南方丘陵山地林道网等级低、通行能力差等问题,挖掘已建路网隐含的工程选线智慧,构建骨干林道建设适宜性评价模型,为林道网扩容升级提供科学依据。方法: 以我国南方8省(区)122个国家森林公园为研究对象,提取3 726.14 km已建林道数据,作为承载工程实践经验的工程案例库。将符合高标准技术规范的主干道界定为骨干道路建设适宜样本,将受地形或需求限制的低等级支路界定为建设约束样本。选取涵盖地形地貌、生态环境和人类活动3个维度的9类特征因子,采用集成学习算法XGBoost建立道路等级分级评价模型,并引入SHAP解释方法,定量解析各因子对骨干林道等级决策的全局贡献和交互补偿效应,最后以中南林业科技大学芦头实验林场为例进行实证评价。结果: 1) XGBoost模型在识别骨干林道与支线林道的建设环境差异上表现优异,AUC值达
中图分类号:
龙晨,王忠伟,王擎天. 基于可解释机器学习的林道网骨干道路建设适宜性评价[J]. 林业科学, 2026, 62(11): 187-196.
Chen Long,Zhongwei Wang,Qingtian Wang. Suitability Evaluation of Backbone Road Construction in Forest Road Networks Based on Explainable Machine Learning[J]. Scientia Silvae Sinicae, 2026, 62(11): 187-196.
表1
林区骨干道路建设适宜性评价因子"
| 影响因素 Influencing Factors | 特征选取 Feature Selection | 精度 Accuracy | 数据来源 Data Source |
| 地形地貌 Topography and Geomorphology | 高程 Elevation | 30 m | 哥白尼数字高程模型 Copernicus digital elevation model |
| 坡度Slope | 30 m | ||
| 地形起伏度Terrain ruggedness index | 30 m | ||
| 生态环境 Ecological Environment | 归一化植被指数 Normalized difference vegetation index(NDVI) | 10 m | Sentinel-2 L2A |
| 土壤侵蚀程度Soil erosion intensity | 1 km | 资源环境科学数据平台 Resource and environmental science data platform | |
| 土壤类型Soil type | 1 km | ||
| 人类活动 Human Activities | 距居民点距离Distance to residential | — | OpenStreetMap |
| 到水系距离Distance to water system | — | ||
| 土地利用类型Land use type | 30 m |
图2
评价因子特征分布 在土壤侵蚀程度图中,11:水力侵蚀微度;12:水力侵蚀轻度;13:水力侵蚀中度;14:水力侵蚀强度;31:冻融侵蚀微度。In the map of soil erosion intensity, 11: Slight water erosion; 12: Light water erosion; 13: Moderate water erosion; 14: Severe water erosion; 31: Slight freeze-thaw erosion. 在主要土壤类型图中,2:棕壤;6:石灰性褐;14:暗棕壤;40:黄壤;44:水稻土;47:石灰(岩)土;50:黄红壤;58:黄棕壤;61:山原红壤;68:红壤;78:赤红壤。In the map of major soil types, 2: Brown earth; 6: Calcareous cinnamon soil; 14: Dark brown earth; 40: Yellow soil; 44: Paddy soil; 47: Limestone soil; 50: Yellow-red soil; 58: Yellow-brown earth; 61: Mountain plateau red soil; 68: Red soil; 78: Lateritic red soil. 在土地利用类型图中,1:耕地;2:林地;3:草地;4:水域;5:城乡、工矿、居民用地;6:未利用土地;8:其他土地. In the map of land use type, 1: Cropland; 2: Forest land; 3: Grassland; 4: Water bodies; 5: Urban and rural residential, industrial and mining land; 6: Unused land; 8: Other land."
表3
基于 5 折交叉验证的 XGBoost 模型分类性能评估"
| 折数 Fold | 准确率 Accuracy | 精确率 Precision | 召回率 Recall | F1-score | AUC面积 area under ROC curve |
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 | |||||
| 5 | |||||
| 平均值Average |
| 黄 钦, 谭 翠, 杨 波. 基于XGBoost算法的亚热带地区生态旅游适宜性评价方法研究. 地球信息科学学报, 2024, 26 (2): 303- 317. | |
| Huang Q, Tan C, Yang B. Research on the evaluation method of ecotourism suitability in subtropical regions based on XGBoost algorithm. Journal of Geo-Information Science, 2024, 26 (2): 303- 317. | |
|
林玉英, 李宝银, 邱荣祖, 等. 基于GWR模型的道路网络对森林碳密度干扰的地理变异: 以闽江上游地区为例. 地理科学, 2022, 42 (6): 1113- 1123.
doi: 10.13249/j.cnki.sgs.2022.06.017 |
|
|
Lin Y Y, Li B Y, Qiu R Z, et al. Geographic variation of road network effects on forest carbon density based on GWR model: a case study of the upstream district of the Minjiang River. Scientia Geographica Sinica, 2022, 42 (6): 1113- 1123.
doi: 10.13249/j.cnki.sgs.2022.06.017 |
|
|
邱荣祖, 吴 强. 山地林道网配置线型与合理密度. 福建林学院学报, 2002, 22 (1): 9- 12.
doi: 10.3969/j.issn.1001-389X.2002.01.003 |
|
|
Qiu R Z, Wu Q. Disposition configuration and optimal density of forest road network. Journal of Fujian College of Forestry, 2002, 22 (1): 9- 12.
doi: 10.3969/j.issn.1001-389X.2002.01.003 |
|
| 谭 翠, 黄 钦, 杨 波, 等. 随机森林算法在区域生态旅游适宜性评价中的应用研究. 地球信息科学学报, 2024, 26 (2): 318- 331. | |
| Tan C, Huang Q, Yang B, et al. Application of random forest algorithm in regional ecotourism suitability assessment. Journal of Geo-Information Science, 2024, 26 (2): 318- 331. | |
|
翁倩文, 陈伟强, 王德勋, 等. 基于集成机器学习的大理红花大金元种植适宜性区划. 中国烟草科学, 2025, 46 (2): 113- 120.
doi: 10.13496/j.issn.1007-5119.2025.02.014 |
|
|
Weng Q W, Chen W Q, Wang D X, et al. Planting suitability zones of honghuadajinyuan in Dali prefecture based on integrated machine learning. Chinese Tobacco Science, 2025, 46 (2): 113- 120.
doi: 10.13496/j.issn.1007-5119.2025.02.014 |
|
| 杨 静, 毛瑾欣, 赵文鑫, 等. 基于单分类机器学习的生活垃圾焚烧发电设施选址可行性与影响因素识别: 来自全国1 058个项目的证据. 中国环境管理, 2025, 17 (6): 28- 38. | |
| Yang J, Mao J X, Zhao W X, et al. One-class machine learning-based feasibility assessment and factor identification for siting waste-to-energy incineration facilities: evidence from 1 058 projects in China. Environmental Conformity Assessment, 2025, 17 (6): 28- 38. | |
|
杨泓宇, 赵建平, 罗 涛. 基于ROC曲线对日光眩光指标等级阈值的研究. 建筑科学, 2025, 41 (4): 156- 164.
doi: 10.13614/j.cnki.11-1962/tu.2025.04.20 |
|
|
Yang H Y, Zhao J P, Luo T. Investigation into the thresholds of daylight glare metric levels based on ROC curve. Building Science, 2025, 41 (4): 156- 164.
doi: 10.13614/j.cnki.11-1962/tu.2025.04.20 |
|
|
张晓瑶, 钟林生. 基于AutoGluon算法的青海省生态旅游适宜性评价. 中国生态旅游, 2024, 14 (6): 1224- 1237.
doi: 10.12342/zgstly.20240305 |
|
|
Zhang X Y, Zhong L S. Evaluation of ecotourism suitability in Qinghai Province based on AutoGluon algorithm. Journal of Chinese Ecotourism, 2024, 14 (6): 1224- 1237.
doi: 10.12342/zgstly.20240305 |
|
|
张玉冰, 申彦波, 姚 鑫, 等. 青海高原光伏适宜性评价的不同决策树算法的比较研究. 太阳能学报, 2024, 45 (12): 30- 39.
doi: 10.19912/j.0254-0096.tynxb.2023-2078 |
|
|
Zhang Y B, Shen Y B, Yao X, et al. Comparative study of different decision tree algorithms for pv suitability evaluation in Qinghai plateau. Acta Energiae Solaris Sinica, 2024, 45 (12): 30- 39.
doi: 10.19912/j.0254-0096.tynxb.2023-2078 |
|
| 周新年, 邱荣祖, 张正雄, 等. 环境友好型的木材物流系统研究进展. 林业科学, 2008, 44 (4): 132- 138. | |
| Zhou X N, Qiu R Z, Zhang Z X, et al. A Review of Eco-Friendly Wood Logistics System Research. Scientia Silvae Sinicae, 2008, 44 (4): 132- 138. | |
|
Abdi E, Majnounian B, Darvishsefat A, et al. A GIS-MCE based model for forest road planning. Journal of Forest Science, 2009, 55 (4): 171- 176.
doi: 10.17221/52/2008-JFS |
|
| Achour Y, Pourghasemi H R. 2020. How do machine learning techniques help in increasing accuracy of landslide susceptibility maps? Geoscience Frontiers, 11(3): 871-883. | |
|
Acosta F C, Rengifo S P, García M L, et al. Road network planning in tropical forests using GIS. Croatian Journal of Forest Engineering, 2023, 44 (1): 153- 169.
doi: 10.5552/crojfe.2023.1742 |
|
|
Aguiar M O, Fernandes da Silva G, Mauri G R, et al. Optimizing forest road planning in a sustainable forest management area in the Brazilian Amazon. Journal of Environmental Management, 2021, 288, 112332.
doi: 10.1016/j.jenvman.2021.112332 |
|
| Caliskan E, Bediroglu S, Yildirim V. 2019. Determination forest road routes via gis-based spatial multi-criterion decision methods. Applied Ecology and Environmental Research, 17(1): 759−779. | |
| Chen T Q, Guestrin C. 2016. XGBoost: a scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. San Francisco California USA. ACM, 785−794. | |
|
Diktas-Bulut N, Gumus S, Karsli H, et al. Determination of rippability classes through classical and geophysical methods in new forest road excavations, and their evaluation in terms of cost and progress payments. Croatian Journal of Forest Engineering, 2025, 46 (2): 369- 385.
doi: 10.5552/crojfe.2025.2488 |
|
|
do Prado Morais P, ARIMA E Y, de Souza Á N, et al. Assessment of forest road models in concession areas in the Brazilian Amazon. Forests, 2023, 14 (7): 1388.
doi: 10.3390/f14071388 |
|
| Duka A, Papa I, Proto A R, et al. 2025. Forest accessibility and road network density: a global overview with a special focus on Europe. Current Forestry Reports, 11(1): 26. | |
|
Engert J E, Campbell M J, Cinner J E, et al. Ghost roads and the destruction of Asia-Pacific tropical forests. Nature, 2024, 629 (8011): 370- 375.
doi: 10.1038/s41586-024-07303-5 |
|
|
Ferreira da Silva E, Fernandes da Silva G, Orfanó Figueiredo E, et al. Optimized forest planning: allocation of log storage yards in the Amazonian sustainable forest management area. Forest Ecology and Management, 2020, 472, 118231.
doi: 10.1016/j.foreco.2020.118231 |
|
|
Genc C O, Akinci H, Kilicoglu C, et al. Assessment of forest road networks for landslide susceptibility: a case study of northern forest area in türkiye. Croatian Journal of Forest Engineering, 2026, 47 (1): 185- 202.
doi: 10.5552/crojfe.2026.3434 |
|
|
Hosseini S A O, Moghadasi P, Fallah A. Forest road network design based on multipurpose forestry management in Hyrcanian forest. Journal of Environmental Science and Management, 2019, 22 (2): 13.
doi: 10.47125/jesam/2019_2/03 |
|
| Huang S Y, Dou H Q, Jian W B, et al. 2023. Spatial prediction of the geological hazard vulnerability of mountain road network using machine learning algorithms. Geomatics, Natural Hazards and Risk, 14: 2170832. | |
| Jahandar O, Abdi E, Jaafari A. 2022. Assessment of slope failure susceptibility along road networks in a forested region, northern Iran. Physics and Chemistry of the Earth, Parts A/B/C, 128: 103272. | |
|
Kadi F. Developing a MATLAB GUI based application for optimal forest road route based on landslide susceptibility maps. Advances in Space Research, 2025, 76 (9): 4858- 4879.
doi: 10.1016/j.asr.2025.07.080 |
|
| Kadi F, Yildirim F, Saralioglu E. Risk analysis of forest roads using landslide susceptibility maps and generation of the optimum forest road route: a case study in Macka, Turkey. Geocarto International, 2021, 36 (14): 1612- 1629. | |
|
Karjalainen V, Malinen J, Tokola T. Predicting bearing capacity and trafficability classes of forest roads using road properties and surrounding terrain information. Road Materials and Pavement Design, 2025, 26 (8): 2037- 2054.
doi: 10.1080/14680629.2024.2434945 |
|
|
Kazama V S, Dalla Corte A P, Robert R C G, et al. Global review on forest road optimization planning: Support for sustainable forest management in Amazonia. Forest Ecology and Management, 2021, 492, 119159.
doi: 10.1016/j.foreco.2021.119159 |
|
|
Kolkos G, Stergiadou A, Kantartzis A, et al. Effects of forest roads and an assessment of their disturbance of the natural enviroment based on GIS spatial multi-criteria analysis: case study of the University Forest of Taxiarchis, Chalkidiki, Greece. Euro-Mediterranean Journal for Environmental Integration, 2023, 8 (2): 425- 440.
doi: 10.1007/s41207-023-00362-6 |
|
|
Lacerda T H S, de Jesus Franca L C, Lopes I L E, et al. Multi-objective forest harvesting under sustainable and economic principles. Journal of Forestry Research, 2023, 34 (5): 1379- 1394.
doi: 10.1007/s11676-023-01614-5 |
|
|
Lin K Y, Zhang L F, Huang L D, et al. Improved particle swarm path planning algorithm with multi-factor coupling in forest fire spread scenarios. Fire, 2023, 6 (5): 202.
doi: 10.3390/fire6050202 |
|
|
Liu Z Q, Gilbert G, Cepeda J M, et al. Modelling of shallow landslides with machine learning algorithms. Geoscience Frontiers, 2021, 12 (1): 385- 393.
doi: 10.1016/j.gsf.2020.04.014 |
|
|
Long C, Pang Y, Wang Z W. Study of the sustainability of a forest road network using GIS-MCE. Forests, 2023, 14 (12): 2410.
doi: 10.3390/f14122410 |
|
| Lu Z L, Sun S F, Yuan M J, et al. Fire path fighting in forest off-road using improved ACA-An example of the northern primitive forest region of the Great Xing’an Range in Inner Mongolia, China. Forests, 2022, 13 (10): 1717. | |
|
Mahpour A, El-Diraby T. Application of machine-learning in network-level road maintenance policy-making: the case of Iran. Expert Systems with Applications, 2022, 191, 116283.
doi: 10.1016/j.eswa.2021.116283 |
|
|
Mesquita M, Marques S, Marques M, et al. An optimization approach to design forest road networks and plan timber transportation. Operational Research, 2022, 22 (3): 2973- 3001.
doi: 10.1007/s12351-021-00640-7 |
|
|
Papa I, Hodak L, Popovic M, et al. ALS and SfM field data survey as a basis of forest road design. Forests, 2026, 17 (2): 265.
doi: 10.3390/f17020265 |
|
| Sakar D, Aydin A, Akay A E. 2020. Using GIS-based multicriteria decision support system for planning road networks with visual quality constraints: a case study of protected areas in Ankara, Turkey. Environmental Monitoring and Assessment, 192(7): 447. | |
|
Sales A, Gonzáles D G E, Martins T G V, et al. Optimization of skid trails and log yards on the Amazon forest. Forests, 2019, 10 (3): 252.
doi: 10.3390/f10030252 |
|
|
Simoes D, Cavalcante F S, Lima R C A, et al. Optimal forest road density as decision-making factor in wood extraction. Forests, 2022, 13 (10): 1703.
doi: 10.3390/f13101703 |
|
|
Solanki A, Gupta V, Joshi M. Application of machine learning algorithms in landslide susceptibility mapping, Kali Valley, Kumaun Himalaya, India. Geocarto International, 2022, 37 (27): 16846- 16871.
doi: 10.1080/10106049.2022.2120546 |
|
| Stefanovic B, Stojnic D, Danilovic M. Multi-criteria forest road network planning in fire-prone environment: a case study in Serbia. Journal of Environmental Planning and Management, 2016, 59 (5): 911- 926. | |
|
Suzuki M, Saito M U. Vehicular traffic frequency and environmental factors affect forest road use by ground-dwelling mammals in northeastern Japan: Management intensity matters. Forest Ecology and Management, 2024, 572, 122318.
doi: 10.1016/j.foreco.2024.122318 |
|
|
Thomas J, Brunette M, Leblois A. The determinants of adapting forest management practices to climate change: lessons from a survey of French private forest owners. Forest Policy and Economics, 2022, 135, 102662.
doi: 10.1016/j.forpol.2021.102662 |
|
|
Veisi O, Moradi M A, Gharaei B, et al. Sustainable forestry logistics: Using modified A-star algorithm for efficient timber transportation route optimization. Forest Policy and Economics, 2025, 173, 103456.
doi: 10.1016/j.forpol.2025.103456 |
| [1] | 焦昱华,张青峰,周传龙,刘金成. 智慧林业背景下森林碳汇估算:进展、挑战与展望[J]. 林业科学, 2026, 62(9): 209-220. |
| [2] | 熊世梅,谭炳香,许文强,李骁尧,庞丽峰,胡冰. 基于无人机多光谱和激光雷达数据的荒漠梭梭林地上生物量估算[J]. 林业科学, 2026, 62(6): 96-108. |
| [3] | 周晟,蒋馥根,陈帅,龙依,王彬彬,宋子戈,孙华. 联合星−空激光雷达和Sentinel-2数据的森林地上生物量估测方法[J]. 林业科学, 2026, 62(3): 74-87. |
| [4] | 施正煌,侯方淼,张浩昆,黄国星. 新一轮集体林权制度改革对中国木质林产品出口边际增长的影响分析[J]. 林业科学, 2025, 61(9): 211-222. |
| [5] | 张宇娇,赵恒谦,付含聪,刘哿,皇甫霞丹,刘轩绮. 基于无人机遥感多特征的单木地上生物量反演模型[J]. 林业科学, 2025, 61(8): 129-141. |
| [6] | 袁泽雨,许行,任怡,许杨,庞建壮,吴小云,张翰遥,张志强. 2001—2021年“三北”工程区植被韧性分布特征及其驱动因素[J]. 林业科学, 2025, 61(7): 182-191. |
| [7] | 杨菲菲,张王菲,赵磊,赵含,姬永杰,王梦金. 基于GF-1数据的二阶段遥感特征优选与森林地上生物量反演模型[J]. 林业科学, 2025, 61(4): 9-19. |
| [8] | 葛晓宁,许新桥,张怀清,张京,杨杰,崔泽宇,傅汝饶,梁金洁,邹添华,王林龙,刘洋. 林木基因型-环境互作算法研究进展与思考[J]. 林业科学, 2025, 61(3): 1-15. |
| [9] | 李倩,张帆,孟祥雪,吴小云,庞建壮,许行,张志强. 基于分解重构与机器学习的华北平原杨树人工林净碳交换模拟[J]. 林业科学, 2025, 61(12): 72-82. |
| [10] | 潘丹,罗璐薏,季凯文,孔凡斌. 农村电子商务发展对林区内部收入差距与共同富裕的影响[J]. 林业科学, 2024, 60(5): 35-50. |
| [11] | 沈琛琛,肖文发,朱建华,曾立雄,陈吉臻,黄志霖. 基于机器学习算法的华中天然林土壤有机碳特征与关键影响因子[J]. 林业科学, 2024, 60(3): 65-77. |
| [12] | 焦强英,韩宗甫,王炜烨,刘迪,潘鹏旭,李博,张念慈,王萍,陶金花,范萌. 基于多源数据和机器学习方法的大兴安岭地区雷击火驱动因子及火险预测模型[J]. 林业科学, 2023, 59(6): 74-87. |
| [13] | 杨楠楠,白岩,姜苏译,杨春梅,徐凯宏. 基于AlexNet优化的板材材种识别方法[J]. 林业科学, 2022, 58(3): 149-158. |
| [14] | 卫玮, 党坤良. 秦岭南坡林地土壤有机碳密度空间分异特征[J]. 林业科学, 2019, 55(5): 11-19. |
| [15] | 厉静文,郭浩,王雨生,辛智鸣,吕永军. 基于MaxEnt模型的胡杨潜在适生区预测[J]. 林业科学, 2019, 55(12): 133-139. |
| 阅读次数 | ||||||
|
全文 |
|
|||||
|
摘要 |
|
|||||